Anyone can check now
On October 6 Google opened its SynthID detection tool to the public at synthid.com. SynthID is the invisible watermarking technology Google introduced in 2023; the detector had previously been limited to select journalists, media professionals and researchers, with testing that began at Google I/O 2025.
Now anyone can upload a file to check: images in JPG, PNG, WEBP, AVIF, HEIC, TIFF, GIF and other formats, video in MP4, MOV and WEBM, and audio in WAV, MP3, OGG, FLAC, AAC and M4A. Detection is also built into the Gemini app and the Chrome browser. Google says it currently handles about 1 million verification requests a day.
Whose content it can detect
Content from Google's own models such as Nano Banana, Veo and Lyria, and tools such as Gemini, Flow, ProducerAI and Vids, carries SynthID. According to TechCrunch, OpenAI, Nvidia and Kakao also support SynthID, with OpenAI running its own verification site as well; Apple is reportedly joining. Microsoft and Meta use their own watermarking and detection tools, and Reuters previously found Meta's detector failing to identify some of Meta's own cropped AI images.
What it means
For the first time, "was this image made by AI?" has an entry point that ordinary people can use and that covers several vendors, which is genuinely useful for newsroom verification and platform moderation. But it answers a narrower question: does the file carry a SynthID watermark? A negative result does not mean something was really photographed; content from models that don't support SynthID, or that has been re-encoded or screenshotted, may not register. Fragmented watermark standards remain the biggest weakness of tools like this. OpenAI's textGrain text watermark, covered here on October 6, is a separate system, and text is not among this site's supported types.